For institutional EdTech businesses to scale, business development must revolve around proving efficacy as a large-scale intervention. The key here is that any government-led regional instalment of EdTech products will be a policy decision and not a purely pedagogical one. One reason pedagogy seems very fragmented and teaching methods differ greatly from teacher to teacher is in academia’s pursuit of novelty — particularly in learning/teaching methodology. When your goal is inventing a new method that is more effective than the baseline method, your research project will need to ensure that the methodology is strictly enforced. For that, you likely need someone who’s already very invested in understanding the methodology — a colleague, or even the very author of the research project. This level of methodological enforcement is very difficult to achieve in the real world, unless you have the budget for extensive training as well as in-term audit. It is also likely that teachers participating in the study exhibit a bias in terms of their general passion in improving their teaching. Simply put, it’s all far too optimistic. ...

As a CEO of an EdTech company and a father of two toddlers, I’ve put a lot of thought into what makes people learn new things and develop to be the person they aspire to be. This isn’t limited to learning that typically happens in schools. Personal growth and development are fundamental requirements for a thriving organisation. I’ve come to observe in my professional career that the amount of learning one makes out of a particular experience differs greatly by person. In sports, for example, the same amount of training can yield drastically different skill levels, depending on what people typically call ‘a knack’ or ‘a talent’. The same phenomenon can be observed in virtually anything. Great programmers acquire new concepts quickly and apply them to their work right away. Great marketers generate customer insight like no one else. Sometimes you’re left in awe of how easily mathematicians spot patterns in a seemingly unrelated set of observations. The list goes on. ...

Nicely covered. Interview in conjunction with Google’s LuanchPad Accelerator Tokyo, which we’ve been selected for Batch One.

Joyz, Inc. (Head office: Shinagawa-ku, Tokyo, Japan; CEO: Yoshiyuki Kakihara; hereinafter referred to as Joyz) announces that it has started joint research with the Saruwatari-Koyama Laboratory, Graduate School of Information Science and Technology, The University of Tokyo, on the application of speech synthesis technology to language acquisition. In this joint research, Joyz will work with Assistant Professor Shinnosuke Takamichi (hereafter, Assistant Professor Takamichi) in Saruwatari Laboratory, Department of Systems Information Science, Graduate School of Information Science and Technology, The University of Tokyo, to develop effective foreign language acquisition methods utilising learning history and speech data owned by Joyz, in an industry-academia collaboration system. ...

I gave a talk at DjangoCongress JP 2018 named “Django in the age of AI - the Good, the Bad and the Ugly”. I have been somewhat distant from the local Python community since I helped run PyCon JP 2014 as a publicity team member. I hope my talk was interesting enough.

AI英会話アプリ「TerraTalk」のHeroku＋AWS活用法 I gave a talk at an AWS user conference. I spoke about the technical challenges of AI English learning app TerraTalk, the back-end configuration and our agility-first architecture. Thank you to everyone involved for the opportunity!